Eliminate Privacy Concerns With Synthetic Data

Eliminate Privacy Concerns with Synthetic Data

Differentially private learning to create fake, synthetic datasets with enhanced privacy guarantees.

[1]:

from gretel_synthetics.config install LocalConfig

config = LocalConfig(
epochs = 15,
gen_lines = 1000,
input_data_path="us-adult-income.csv"
)

[2]:

from gretel_synthetics.train import train_rnn

‍train_rnn(config)

[3]:

from gretel_synthetics.generate import generate_text

for line in generate_text(config):
print(line)

Synthetic results

Age	Gender	Education	Occupation
58	Female	Some college	Adm-clerical
10	Female	Associates	Protective-serv
12	Male	Associates	Tech-support
37	Male	Some college	Sales
35	Female	Bachelors	Farming-fishing
72	Female	HS Grad	Craft-repair
21	Male	Bachelors	Tech-support
54	Male	Some college	Adm-clerical
37	Male	Doctorate	Prof-specialty
44	Female	Some college	Adm-clerical
16	Female	Doctorate	Sales
44	Male	HS Graduate	Sales
39	Female	Some college	Engineering
8	Male	Associates	Adm-clerical
24	Male	Associates	Engineering
58	Female	Some college	Adm-clerical
10	Female	Associates	Protective-serv
12	Male	Associates	Tech-support
37	Male	Some college	Sales
35	Female	Bachelors	Farming-fishing
72	Female	HS Grad	Craft-repair
21	Male	Bachelors	Tech-support
54	Male	Some college	Adm-clerical
37	Male	Doctorate	Prof-specialty
44	Female	Some college	Adm-clerical
16	Female	Doctorate	Sales
44	Male	HS Graduate	Sales
39	Female	Some college	Engineering
8	Male	Associates	Adm-clerical
24	Non-binary	Associates	Engineering

Check out our GitHub for research, source code and examples including our core synthetic data generation library.

gretelai / gretel-synthetics

Star

Create safe data with synthetic datasets easily.

Generate synthetic datasets that retain the same insights and are statistically equivalent to your original data source.

Walkthrough:

Build privacy guarantees
into your existing workflows.

Create models with large amounts of artificial data that generalize better than those trained on limited datasets, with the added benefit of pretecting your customers’ privacy.

Start for free

Improve limited datasets with synthetic data

Use synthetic data to augment data sources, improve accuracy, and reduce bias in machine learning models.

Read the case study

Create synthetic data with privacy guarantees

Create and share realistic synthetic data freely across teams and organizations with differential privacy guarantees

Read the case study

Jumpstart

Get started quickly with Gretel Blueprints

For more advanced usage, we have created a collection of Blueprints to help jumpstart your transformation workflows.

Get a free API key

Synthetic data for everyone

Data Scientists

Build and generate models that are mathematically guaranteed to be free of PII.

ML Engineers

Train AI models on synthetic data without worrying about exposing PII or other sensitive data.

Data Engineers

Anonymize precise customer data and share it safely across teams.

Product engineers

Make datasets publicly available so that engineers can monetize tools built for your data.

Train machine learning models

Generate synthetic data to augment your datasets. This can help you create AI and ML models that perform and generalize better, while reducing algorithmic bias.

Seamlessly share data

No need to snapshot production databases to share with your team. Define transformations to your data with software, and invite team members to subscribe to data feeds in real-time

Privacy, applied

Apply state of the art NLP processing to label personal data and PII in your data streams. Stay compliant by encrypting records containing unexpected PII in real-time.

Gretel Blog

Latest posts about Synthetic Data

See all posts

Introducing Gretel's Privacy Filters

Create synthetic data that’s safer than ever. Our simple configuration file settings enable you to secure both your data and model from adversarial attacks.

Build a synthetic data pipeline using Gretel and Apache Airflow

In this blog post, we build an ETL pipeline that generates synthetic data from a PostgreSQL database using Gretel’s Synthetic Data APIs and Apache Airflow.

Walkthrough: Create Synthetic Data from any DataFrame or CSV

Train an AI model to create an anonymized version of your dataset using Python, Pandas, and gretel-synthetics.

Synthetic data